Why FTP Doesn’t Win Bike Races: What Milan–San Remo Reveals About Pogačar
- Jake Hollins
- Jan 21
- 6 min read
Bike races, like any race, come down to one thing: who crosses the line first. Everyone starts together. To finish first, you have to get from A to B faster than everyone else. So stripped back to basics, how to win a bike race is about understanding speed – where it comes from, what limits it, and how it’s actually used in racing.
Cycling power vs speed: what actually determines how fast you go
At its most basic level: Speed = rider-generated power / resistance (Gravity × rolling resistance × aerodynamic drag)
It’s not quite that simple, because resistance increases as speed increases – meaning more power is required not just to accelerate, but to hold higher speeds. So a more accurate way to think about it is:
Power required = Speed × Total resistive force Total resistive force = Gravity + Rolling resistance + Aerodynamic drag
A very clever Google engineer created this webpage to help you (and me) understand this concept: https://www.gribble.org/cycling/power_v_speed.html
What does riding faster actually mean?
There are only two real options if you want to understand how to win a bike race:
Pedal harder (increase power) Increase the engine through training, nutrition, and recovery.
Reduce resistance or increase efficiency Improve bike handling and skills, drafting and bunch positioning, optimise equipment, and refine riding position.
Every training decision, equipment choice, or bike racing strategy ultimately pulls on one of these ideas.
That makes it useful to think in terms of three performance levers.
Bike racing strategy: the three levers that actually decide races
Engine
How much power you can produce, and how often you can repeat it. This is what most riders focus on. Training, recovery, and fueling determine the ceiling of what’s possible. Without an engine, nothing else matters, but an engine alone rarely wins races.
Efficiency
How much speed you get for the power you spend. This includes:
aerodynamics and riding position
rolling resistance and tire choice
braking, cornering, and bike handling
avoiding unnecessary surges
Two riders producing the same power can be travelling at very different speeds depending on how efficiently that power is converted into forward motion.
Decisions
When and where you choose to spend your limited power.
This is where cycling race strategy becomes decisive. Racing forces choices: when to expose yourself, when to stay sheltered, when to commit. Spending even a small amount of power at the wrong moment can be far more costly than spending a large amount at the right one.
A useful way to frame it is:
‘Better decisions usually mean you get to spend your engine when resistance is lowest – or force others to spend theirs when resistance is highest.’
This is why the strongest rider doesn’t always win.
Pogacar on the Col de Rates: a clear example of the levers in action
Most of you will have seen Tadej Pogačar’s recent, fairly outrageous KOM on the Col de Rates, where he smashed his own already ridiculous record.
This is a great example, not because it was dramatic, but because it allows us to separate power, efficiency, and decisions in a fairly controlled way.
Using this spreadsheet inspired by Gribble’s model, we can estimate the power required for both efforts and then pull the gravity, rolling, and aerodynamic drag levers one by one to see how each affects the result.
Baseline assumptions
Total mass (bike + rider): 72 kg
CdA: 0.27 (a good pro-level aero position)
Rolling resistance (Crr): 0.004
Air density: 1.226 kg/m³
Drivetrain loss: 2%
Headwind: 5 km/h
Gradient: 5.5%
Distance: 6.5 km
Pogacar’s Col de Rates baseline effort – December 2024 (12 min 21 s)
12 min 21 s
Total: ~508 W
Wind drag: ~146 W
Rolling resistance: ~21 W
Gravity: ~330 W
Drivetrain losses: ~9 W
This assumes no drafting effects, which wasn’t the case. He had riders in front of him for roughly 65% of the climb, giving around a 40% reduction in aerodynamic drag. This is us pulling the reduce-resistance lever. Same time, reduced resistance So here we are turning the reduce resistance lever and we see the power needed goes down dramatically.
12 min 21 s Total: ~472 W
Wind drag: ~111 W (–35 W)
Rolling resistance: ~21 W
Gravity: ~330 W
Drivetrain losses: ~9 W
Nothing else has changed. The engine is smaller, but the outcome is identical.
On the 19th of December 2025, Pogacar went back and smashed his Col de Rates record by 25 seconds.
Where did the extra 25 seconds come from? The obvious answer is more power, assuming nothing changes other than increased wind resistance from the higher speed.
11 min 57 s
Total: ~495 W (+23 W)
Wind drag: ~122 W (+11 W)
Rolling resistance: ~22 W
Gravity: ~341 W
Drivetrain losses: ~9 W
But that isn’t the only plausible explanation:
Efficiency improvements
We could also assume:
he’s roughly 1 kg lighter
rolling resistance is marginally lower
his new aero bike reduces frontal area by around 19%
With those changes:
11 min 57 s
Total: ~484 W
Wind drag: ~117 W (–5 W)
Rolling resistance: ~20 W (–1 W)
Gravity: ~336 W (–5 W)
Drivetrain losses: ~9 W
Better decisions
Now assume he isn’t actually stronger than last year. Could he still go 25 seconds faster?
What if the difference actually came from a better team lead-out, a better pacing plan, and more time spent sheltered before committing fully.
In practice, that might simply mean someone else does a bit more work. For example, if it’s Isaac del Toro who finds an extra ~25 W, or covers an additional kilometre on the front, Pogacar benefits from that same ~40% drag reduction for six kilometres instead of five.
11 min 57 s Total: ~471 W
Wind drag: ~105 W (–17 W)
Rolling resistance: ~20 W (–2 W)
Gravity: ~336 W (–5 W)
Drivetrain losses: ~9 W
Why is racing harder to model than climbing?
There’s one caveat: climbs are easier to model because they’re close to steady-state efforts. We can model lead-outs simply, but not the constantly shifting dynamics of a peloton.
That said, the principles don’t change. The underlying levers stay the same, racing just punishes poor decisions faster.
Tadej Pogacar and Milan–San Remo
This framework helps explain why Tadej Pogacar winning Milan-San Remo remains such an unresolved question.
I think if Pogacar could win any race in 2026, it would be this one. To do it, he almost certainly needs to arrive at the finish alone. But 2025 showed that he can’t simply ride everyone off his wheel and “see you in the douches”.
If Mathieu van der Poel can match a given power while sheltered, then Pogacar has to exceed that power by roughly the follower’s aerodynamic saving just to create separation.
The Poggio
At roughly 4%, the Poggio isn’t steep enough for gravity to dominate.
If MVDP can hold ~500 W for 5.5 minutes after 250 km of racing, Pogacar would likely need closer to 530 W to create meaningful separation. That’s an extremely narrow margin.
The Cipressa
The Cipressa is longer, comes earlier, and is tactically more interesting.
This climb comes with ~30 minutes / 25 km left to race. From 2025, we can estimate MVDP was producing around ~480 W in the wheels, while Pogacar likely did something similar while attacking partway up.
What if he waited another km to attack, could he then get a gap and make it stick?
To really put MVDP under pressure, Pogacar likely needs to see ~500 W to the top. That means UAE riding 515–550 W turns on the front. For the final rider, that’s roughly six minutes at ~480 W, followed by a minute at 500–550 W – similar to what they produced on the Col de Rates.
Now all this is assuming no wind. Add in a headwind and it’s even harder for UAE and Pogacar, as the draft benefit increases even more being on the wheel. A strong tailwind swings things back in Pogacar’s favour, as drafting benefits shrink significantly above ~20–25 km/h.
Even then, he’d still need to sustain ~380–400 W for 30 minutes to hold ~51 km/h and hold off the bunch. And if you’re in a small chasing group with MVDP, are you really going to give him a turn for him just to pump you in the sprint?
After writing this, I wouldn’t put Pogacar as the outright favourite – but I wouldn’t bet against him either.
How much power is actually needed to hit a target time?
If you want to explore this beyond theory, or test it against your own climbs, the spreadsheet below lets you do exactly that.
The model is built so that a user can do one simple thing: enter their current time, enter a target time, then move the levers and watch the required power and the power breakdown change.
How it works:
Enter course and rider assumptions
Enter your current time and target time
Adjust key decisions
drafted distance fraction
draft reduction
pacing settings
Read the output
total power required
how that power splits between gravity, rolling resistance and aerodynamic drag
If you’d like to try this on your own climb, you can access the sheet here:
Comments